CN111782957A - Method and device for recommending content, computing equipment and storage medium - Google Patents

Method and device for recommending content, computing equipment and storage medium Download PDF

Info

Publication number
CN111782957A
CN111782957A CN202010671209.6A CN202010671209A CN111782957A CN 111782957 A CN111782957 A CN 111782957A CN 202010671209 A CN202010671209 A CN 202010671209A CN 111782957 A CN111782957 A CN 111782957A
Authority
CN
China
Prior art keywords
content
recommended
contents
candidate
recommendation
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202010671209.6A
Other languages
Chinese (zh)
Inventor
骆顺昌
吴德龙
邱泰生
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tencent Technology Shenzhen Co Ltd
Original Assignee
Tencent Technology Shenzhen Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Tencent Technology Shenzhen Co Ltd filed Critical Tencent Technology Shenzhen Co Ltd
Priority to CN202010671209.6A priority Critical patent/CN111782957A/en
Publication of CN111782957A publication Critical patent/CN111782957A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9535—Search customisation based on user profiles and personalisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/906—Clustering; Classification
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"

Abstract

The application discloses a method, a device, a computing device and a storage medium for recommending contents, which are used for improving the diversity and effectiveness of recommended contents and reducing recommendation redundancy. The method comprises the following steps: acquiring at least two categories of contents to be recommended of a target user; performing multiple rounds of selection on the contents to be recommended of at least two categories according to the recommendation contribution prediction value, and determining the recommendation sequence of the contents selected in each round according to the selection rounds until all the contents to be recommended are sequenced; in each round of selection: respectively determining a recommended contribution predicted value of each candidate content in the current round relative to the category set to which the candidate content belongs, determining at least one candidate content of which the recommended contribution predicted value meets a set condition from all the candidate contents in the current round as the selected content in the current round, removing the selected content from the candidate content, and correspondingly adding the selected content into the category set to which the candidate content belongs; and recommending the contents to the target user according to the sequence of the contents to be recommended which are arranged according to the finished recommendation sequence.

Description

Method and device for recommending content, computing equipment and storage medium
Technical Field
The present application relates to the field of computer technologies, and in particular, to a method and an apparatus for recommending content, a computing device, and a storage medium.
Background
With the rapid development of the internet, users can more and more conveniently contact with massive contents, but information overload is easily caused, for example, in a recommendation scene, for example, in scenes such as information recommendation, short video recommendation and the like, the contents to be recommended are more, and if the contents are not recommended redundantly, the satisfaction degree of the users is lower.
Therefore, the recommendation system can filter the content and then recommend the content to the user, in the related art, the user is generally recommended in a matching mode according to the preference of the user, for example, the user likes to watch the makeup video, and the recommendation system recommends a large number of makeup videos for the user.
Therefore, how to reduce the recommendation redundancy of the recommended content is a problem to be considered.
Disclosure of Invention
The embodiment of the application provides a method, a device, a computing device and a storage medium for recommending contents, which are used for improving the diversity and effectiveness of recommended contents of a recommending system, reducing recommendation redundancy and improving the recommendation performance of the recommending system.
In one aspect, a method of recommending content is provided, the method comprising:
acquiring at least two categories of contents to be recommended of a target user, wherein each category comprises at least one content to be recommended;
performing multiple rounds of selection on the contents to be recommended of the at least two categories according to the recommendation contribution prediction value, and determining the recommendation sequence of the contents selected in each round according to the selection rounds until all the contents to be recommended are sequenced; wherein, in each round of selection: respectively determining a recommended contribution predicted value of each candidate content in the current round relative to the category set to which the candidate content belongs, determining at least one candidate content of which the recommended contribution predicted value meets a set condition from all the candidate contents in the current round as the content selected in the current round, and removing the selected content from the candidate content and correspondingly adding the selected content into the category set to which the candidate content belongs; wherein the candidate content in each round of selection is not selected before the round of selection;
and recommending the contents to the target user according to the sequence of the contents to be recommended which are arranged in the recommendation sequence.
In one aspect, an apparatus for recommending content is provided, the apparatus comprising:
the system comprises an acquisition module, a recommendation module and a recommendation module, wherein the acquisition module is used for acquiring at least two categories of contents to be recommended of a target user, and each category comprises at least one content to be recommended;
the sequencing module is used for carrying out multiple rounds of selection on the contents to be recommended of the at least two categories according to the recommendation contribution prediction value, and determining the recommendation sequence of each round of selected contents according to the selection rounds until all the contents to be recommended are sequenced; wherein, in each round of selection: respectively determining a recommended contribution predicted value of each candidate content in the current round relative to the category set to which the candidate content belongs, determining at least one candidate content of which the recommended contribution predicted value meets a set condition from all the candidate contents in the current round as the content selected in the current round, removing the selected content from the candidate content and correspondingly adding the selected content into the category set to which the selected content belongs, wherein the candidate content in each round of selection is not selected before the round of selection;
and the recommending module is used for recommending the content to the target user according to the sequence of the contents to be recommended which are arranged in the recommending sequence.
Optionally, the obtaining module is configured to:
acquiring a to-be-recommended list corresponding to the target user, wherein the to-be-recommended content in the to-be-recommended list is recalled according to user characteristic data of the target user, and the recommendation sequence of the to-be-recommended content in the to-be-recommended list is related to the matching degree between each to-be-recommended content and the user characteristic data;
and selecting at least two categories of contents to be recommended from the contents to be recommended with high matching degree according to the matching degree of each content to be recommended in the list to be recommended.
Optionally, the sorting module is configured to:
and selecting the candidate content with the largest recommendation contribution predicted value from all the candidate contents in the current round as the selected content in the current round.
Optionally, the sorting module is configured to:
determining a preset number of candidate contents from the maximum recommended contribution predicted value according to the descending order of the recommended contribution predicted value, wherein the preset number is an integer greater than or equal to 2;
determining the content selected in the current round from the predetermined number of candidate contents.
Optionally, the sorting module is configured to:
and according to the sequence of the recommended contribution predicted values from large to small, if the difference value between the recommended contribution predicted values of every two adjacent candidate contents in the predetermined number of candidate contents is smaller than the difference threshold value, determining the selected contents in the current round from the predetermined number of candidate contents.
Optionally, the sorting module is configured to:
determining all the candidate contents in the preset number as the contents selected in the current round; alternatively, the first and second electrodes may be,
and selecting candidate contents meeting the priority ordering condition from the preset number of candidate contents as the contents selected in the current round.
Optionally, the sorting module is configured to:
taking the candidate content with the category different from that of the content selected in the previous round in the preset number of candidate contents as the content selected in the current round; alternatively, the first and second electrodes may be,
and determining a target category of which the number of the selected contents included in the at least two categories before the current round of selection meets a set number limit, and taking the candidate contents belonging to the target category in the predetermined number of candidate contents as the selected contents in the current round.
Optionally, the sorting module is configured to:
and taking the candidate content with the highest original ranking of the candidate contents in the preset number as the content selected in the current round, wherein the original ranking of the candidate contents is ranking according to the sequence from high to low of the matching degree between the candidate contents and the user characteristic data of the target user.
Optionally, the sorting module is configured to:
calculating a first total recommendation contribution prediction value of all selected contents in the category set to which each candidate content belongs through a submodular function;
calculating a second total recommendation contribution prediction value of each candidate content and all selected contents in the category set to which the candidate content belongs through the secondary modulus function;
and determining a recommendation contribution prediction increment of the second total recommendation contribution prediction value relative to the first total recommendation contribution prediction value of each candidate content as the recommendation contribution prediction value of the candidate content.
Optionally, the sorting module is configured to:
the recommendation order is arranged for the contents selected in each round in such a way that the recommendation order of the contents corresponding to the selection is earlier in the selection round.
In one aspect, a computing device is provided, which includes a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes the computer program, the method for recommending content includes the steps.
In one aspect, a storage medium is provided, the storage medium storing computer-executable instructions for causing a computer to perform the steps included in the method for recommending content described above.
In one aspect, a computer program product containing instructions is provided, which when run on a computer causes the computer to perform the steps included in the method of recommending content described in the various possible implementations described above.
In the embodiment of the application, the contents to be recommended of multiple categories are obtained, then, multiple rounds of selection are carried out on all the contents to be recommended according to the recommendation contribution prediction values of the contents to be recommended, the recommendation sequence is arranged for the selected contents selected in each round according to the selection rounds until all the contents to be recommended are sequenced, and further, the contents are recommended to the user according to the sequencing of all the contents to be recommended which are sequenced.
In each round of selection, the recommendation contribution predicted value of each candidate content in the round relative to the category set to which the candidate content belongs is determined, and then one or more candidate contents with the recommendation contribution predicted values meeting the set conditions are selected from all the candidate contents in the round as the contents selected in the round to participate in the sequencing, so that the contents meeting the set conditions relative to the recommendation contribution predicted values of the category set to which the candidate content belongs are selected from the contents to be recommended in all the categories to be sequenced in priority, namely, the contents with high recommendation benefits are ranked in front, and therefore the total recommendation benefits of all the contents to be recommended are ensured to be the maximum as much as possible. Meanwhile, because multiple rounds of selection are carried out, each round of selection determines each round of selection content according to the recommendation income prediction value of each content to be recommended, which is equivalent to scattering and sequencing all types of content, thereby achieving the purpose of suppressing the quantity of the same type of content, enabling the multiple types of content to be arranged at intervals in a cross way as far as possible, increasing the variety of the recommended content, and achieving the purpose of maximizing the total recommendation income of all the contents to be recommended.
After contents to be recommended are reordered, the contents of multiple types are scattered as much as possible and then are arranged in a cross mode, the contents of the multiple types can be recommended in the process of sequentially recommending according to the reordered recommending sequence, the purpose of improving the diversity of the recommended contents is achieved, and therefore the surprising and surprised recommending experience can be provided for users, the potential favor of the users can be mined to a certain degree, the viscosity of the users is further improved, and the retention rate of the users is improved. In addition, the recommendation contribution value of the content which is more advanced is relatively larger for all the contents to be recommended, so that the sequential recommendation can also ensure that the overall recommendation income of all recommended contents to be recommended is maximized as much as possible, thereby ensuring the accuracy and effectiveness of the recommendation.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, it is obvious that the drawings in the following description are only the embodiments of the present application, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
Fig. 1 is a schematic diagram of an application scenario applicable to the embodiment of the present application;
FIG. 2 is a flowchart illustrating a method for recommending content according to an embodiment of the present application;
fig. 3 is a schematic diagram of at least two categories of acquired content to be recommended in an embodiment of the present application;
FIG. 4 is a schematic illustration of a first round of selection in an embodiment of the present application;
FIG. 5a is a schematic diagram illustrating the selection of candidate content with a recommended contribution prediction increment satisfying a set condition as selected content in an embodiment of the present application;
FIG. 5b is another diagram illustrating the selection of a candidate content with a recommended contribution prediction increment satisfying a set condition as a selected content in the embodiment of the present application;
FIG. 5c is another diagram illustrating the selection of a candidate content with a recommended contribution prediction increment satisfying a set condition as a selected content in the embodiment of the present application;
FIG. 6 is a schematic illustration of a diminishing marginal benefit;
FIG. 7 is a schematic flow chart illustrating a method for recommending content in an embodiment of the present application;
fig. 8 is a schematic structural diagram of an apparatus for recommending content in an embodiment of the present application;
fig. 9 is a schematic structural diagram of a computing device in an embodiment of the present application.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the technical solutions in the embodiments of the present application will be described clearly and completely with reference to the accompanying drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by one of ordinary skill in the art from the embodiments given herein without making any creative effort, shall fall within the scope of the claimed protection. In the present application, the embodiments and features of the embodiments may be arbitrarily combined with each other without conflict. Also, while a logical order is shown in the flow diagrams, in some cases, the steps shown or described may be performed in an order different than here.
The terms "first" and "second" in the description and claims of the present application and the above-described drawings are used for distinguishing between different objects and not for describing a particular order. Furthermore, the term "comprises" and any variations thereof, which are intended to cover non-exclusive protection. For example, a process, method, system, article, or apparatus that comprises a list of steps or elements is not limited to only those steps or elements listed, but may alternatively include other steps or elements not listed, or inherent to such process, method, article, or apparatus. The "plurality" in the present application may mean at least two, for example, two, three or more, and the embodiments of the present application are not limited.
In addition, the term "and/or" herein is only one kind of association relationship describing an associated object, and means that there may be three kinds of relationships, for example, a and/or B, which may mean: a exists alone, A and B exist simultaneously, and B exists alone. In addition, the character "/" in this document generally indicates that the preceding and following related objects are in an "or" relationship unless otherwise specified.
Some terms referred to herein are explained below to facilitate understanding by those skilled in the art.
1. The recommendation system is an information filtering system, comprises a large amount of contents to be recommended, performs content filtering and screening aiming at each user, and can recommend proper contents for the corresponding user. The recommendation system may include, for example, an information recommendation system, a short video recommendation system, a music recommendation system, and the like, according to the recommended content.
2. The content in the embodiment of the present application mainly refers to content recommended to the user through the recommendation system, and may include information, short videos, music, news, books, applications, and other types.
3. The marginal benefit is decreased, that is, if an input element is continuously and equivalently increased under the condition that other conditions are not changed, the increment of the provided product is reduced after the input element is increased to a certain output value, namely, the marginal output of the variable element is decreased. As consumers consume an increasing total quantity of an item, the resulting utility (i.e., marginal utility) of the newly added last unit item generally exhibits a decreasing phenomenon (decrementing) referred to as a marginal benefit decrementing, or referred to as a marginal decrementing effect, or referred to as a marginal utility decrementing.
Popular understanding: at the beginning, the profit value is high, and later, the profit value is less, expressed in mathematical language: x is an independent variable, y is a dependent variable, y changes along with the change of x, and the increasing amount of y is continuously reduced along with the increase of the value of x.
4. The submodel function is also called sub-submodel function or submodel function, the submodel function has submodel, and the submodel is also called sub-submodel or submodel, and is a formal description of the marginal benefit decreasing phenomenon, so the submodel function is suitable for the description and the calculation of participating in the marginal benefit decreasing task.
The idea of the present application is presented below.
As described above, most of recommendation systems in related technologies concern how to improve recommendation accuracy when recommending content for a user, in these recommendation mechanisms, generally, matching recommendation is performed on the user according to user preferences, and in the past, the recommendation method recommends a single content, which is easy to form an information cocoon room, and as a result, recommended contents of the same user are classified to be the same or similar, and are similar and redundant in content, which is easy to cause user annoyance, that is, the related technologies ignore diversity of recommendation results when recommending content, and have the problem of recommendation redundancy, and the recommendation effectiveness is low.
Specifically, the recommendation system in the related art generally only focuses on recalls and ranking in the recommendation process, and how to perform diversity enhancement to solve the problem of recommendation redundancy of recommended content is not processed. For example, after receiving a recommendation request from a user, the recommendation system may recall some contents to be recommended from a recommendation pool including a large amount of contents according to user data of the user, that is, the recalled contents to be recommended are adapted to the user data of the user, which is simply understood as being liked by the user approximately, and further, perform scoring and sorting according to a matching degree between each recalled content to be recommended and the user data of the user. The probability of clicking each content to be recommended by the user can be calculated according to user data of the user, the probability of clicking the content can be referred to as click rate for short, the click rate can represent the likeness degree of the content to be recommended by the user, after the click rate of each content to be recommended is obtained, each content to be recommended can be scored according to the click rate to obtain the score value of each content to be recommended, all the contents to be recommended are ranked according to the order of the score values from high to low to obtain a final list to be recommended, the higher the score value is, the higher the probability that the user clicks the corresponding content to be recommended is, and therefore when the content is recommended for the user, the content can be recommended to the user sequentially according to the order of the score values from high to low.
For example, the label of the user is "information like entertainment type", based on the recommendation scheme in the related art, the score value is higher, that is, most of the information ranked before is entertainment type information, if the information recommendation is performed according to the order of the score value from high to low, most of the information recommended to the user is entertainment type information, and multiple pieces of information content of the same type (that is, entertainment type) may have content duplication, so that the user is pushed the same type of information for a long time, lacks novelty and surprise, may cause the user to be bored, may cause the interest of the user to be attenuated, and may gradually decrease the viscosity of the recommendation system.
In view of this, the inventor of the present application considers that a novel and surprising recommendation experience can be provided to a user to some extent by increasing the diversity of the recommendation content, according to the principle of diminishing marginal benefit, in each recommendation, the more the same type of item in the recommended item (e.g. information or short video), the more the total profit is, but the more the same type of item is increased, the less the profit increase is, that is, as the number of the recommended same type of item is greater, the more the total profit is increased, but as the number of the same type of item is greater, the less the total profit increase (i.e. profit increase) is, that is, the more the same type of item is, the less the total profit increase trend of the same type of item is. Therefore, in order to maximize the total profit, the profit increment needs to be increased, which requires that the same type of article should not be recommended too much, and the diversity of the article type needs to be increased, so that the profit increase is accelerated by the diversity of the article type, that is, the profit increment is increased as much as possible by increasing the diversity of the article type, thereby maximizing the total profit.
The above principle of marginal benefit decrement is applied to the recommendation scene in the embodiment of the present application, in order to maximize the total recommendation benefit of the recommendation system, the total recommendation benefit here may be understood as the total click rate of the user on all recommended contents, that is, the total interested value of all recommended contents is quantified, and the contents with high recommendation benefit need to be arranged in front, and the number of the same kind of contents needs to be suppressed. Based on this, the recommended scheme in the examples of the present application is proposed as follows: the method comprises the steps of firstly obtaining a plurality of categories of contents to be recommended, then carrying out a plurality of rounds of selection on all the contents to be recommended according to the recommendation contribution prediction values of all the contents to be recommended, arranging a recommendation sequence for the selected contents selected in each round according to the selection rounds until all the contents to be recommended are sequenced, and further recommending the contents to a user according to the sequencing of all the contents to be recommended which are arranged according to the completed recommendation sequence. In each round of selection, firstly determining a recommendation contribution predicted value of each candidate content in the round relative to a category set to which the candidate content belongs, and then selecting one or more candidate contents of which the recommendation contribution predicted values meet set conditions from all the candidate contents in the round as the contents selected in the round to participate in the sorting, so that the contents with high recommendation income are sorted by priority, namely the contents with high recommendation income are sorted in front, from the contents to be recommended of all the categories, wherein the recommendation contribution predicted values of the category set to which the candidate content per se belongs meet the set conditions (for example, the recommendation contribution predicted values are maximum), thereby ensuring that the total recommendation income of all the contents to be recommended to a user is maximum by a recommendation system as much as possible, and simultaneously, because of multiple rounds of selection, each round of selection determines the selected contents in each round according to the recommendation contribution predicted values of the contents to be recommended, so that the contents of all the categories are sorted by scattering, therefore, the purpose of suppressing the number of the same type of content is achieved, the content of multiple types is arranged in a cross interval mode as scattered as possible, the type diversity of the recommended content is increased, and the purpose of maximizing the total recommendation income of all the content to be recommended is achieved.
After the contents to be recommended are reordered through the scheme, the contents of multiple types are scattered as much as possible and then are arranged in a cross mode, in the process of sequentially recommending according to the reordered recommendation sequence, the contents of multiple types can be recommended, the purpose of improving the diversity of the recommended contents is achieved, and therefore the surprising and surprised recommendation experience can be provided for a user, the potential favor of the user can be mined to a certain extent, the viscosity of the user is further improved, and the retention rate of the user is improved. In addition, the recommendation contribution value of the content which is more advanced is relatively larger for all the contents to be recommended, so that the sequential recommendation can also ensure that the overall recommendation income of all recommended contents to be recommended is maximized as much as possible, thereby ensuring the accuracy and effectiveness of the recommendation.
It should be noted that the diversity referred to in the embodiments of the present application refers to the diversity of categories to which recommended content belongs, for example, content with a top recommendation order belongs to various categories, so that it is ensured that various types of content are recommended to a user; and, the diversity designed in the embodiment of the present application also refers to the diversity of category distribution when performing content recommendation, i.e. the content of multiple categories is arranged alternately, for example, the entertainment information is ranked 1 st and 2 nd in the recommendation sequence, the military information is ranked 3 rd, the scientific information is ranked 4 th and 5 th, the entertainment information is ranked 6 th and 7 th, the scientific information is ranked 8 th, and so on. Therefore, the diversity recommendation of the content is improved, and the accuracy and the effectiveness of the recommendation can be ensured.
The method for recommending the content can be applied to various recommendation scenes, such as a scene for recommending videos, a scene for recommending information, a scene for recommending electronic books, a scene for recommending applications and the like.
In order to better understand the technical solution provided by the embodiment of the present application, some brief descriptions are provided below for application scenarios to which the technical solution provided by the embodiment of the present application is applicable, and it should be noted that the application scenarios described below are only used for illustrating the embodiment of the present application and are not limited. In specific implementation, the technical scheme provided by the embodiment of the application can be flexibly applied according to actual needs.
Referring to fig. 1, fig. 1 is an application scenario in which the method for recommending content in the embodiment of the present application is applied, and the method for recommending content may be applied to a content recommendation system. As shown in fig. 1, the application scenario includes a plurality of terminal devices (e.g., terminal device 101, terminal device 102, terminal device 103) and a server 104, where the server 104 may be a server serving a content recommendation platform, such as an information recommendation server, a short video recommendation server, and so on. Each terminal device is in communication connection with the server 104, and each terminal device may send content to the server 104 to be published to a content recommendation platform served by the server 104, and may also receive published content that has been published to the content recommendation platform and is sent by the server 104.
Taking the terminal device 102 and the short video recommendation scene as an example, the terminal device 102 corresponds to the user 2, and the user 2 may operate the terminal device 102 to send a short video recommendation request to the server 104, further, the server 104 may perform diversity ranking on the short videos to be recommended by using the content recommendation method in the embodiment of the present application, and then recommend the short videos to the terminal device 102 for the user 2 to watch according to the recommendation sequence of the short videos sorted by the diversity ranking, so that the user 2 can look up various types of short videos as much as possible, and novelty is increased, thus, the potential preference of the user 2 can be mined to a certain extent, and the surprise degree of content recommendation is increased.
Taking the terminal device 101 and the information recommendation scenario as an example, the terminal device 101 corresponds to the user 1, the user 1 may operate the terminal device 101 to send an information recommendation request to the server 104, further, the server 104 may perform diversity ranking on the information to be recommended by using the content recommendation method in the embodiment of the present application, and then recommend the information to the terminal device 101 for the user 1 to watch according to the recommendation sequence of the information after diversity ranking, so that the user 1 can check various types of information as much as possible, and the novelty is increased, thus, the potential preference of the user 1 can be mined to a certain extent, and the surprise of content recommendation is increased.
The server 104 may be an independent physical server, a server cluster or a distributed system formed by a plurality of physical servers, or a cloud server providing basic cloud computing services such as a cloud service, a cloud database, cloud computing, a cloud function, cloud storage, a Network service, cloud communication, a middleware service, a domain name service, a security service, a CDN (Content Delivery Network), a big data and artificial intelligence platform, and the like. The terminal devices 101, 102, 103 may be, but are not limited to, a smart phone, a tablet computer, a notebook computer, a desktop computer, a smart television, a smart wearable device, and the like.
To further illustrate the technical solutions provided by the embodiments of the present application, the following detailed description is made with reference to the accompanying drawings and the detailed description. Although the embodiments of the present application provide the method operation steps as shown in the following embodiments or figures, more or less operation steps may be included in the method based on the conventional or non-inventive labor. In steps where no necessary causal relationship exists logically, the order of execution of the steps is not limited to that provided by the embodiments of the present application. The method can be executed in sequence or in parallel according to the method shown in the embodiment or the figure when the method is executed in an actual processing procedure or a device.
The present application provides a method of recommending content that may be performed by a recommendation server in a recommendation system, such as server 104 in fig. 1. The method for recommending content provided by the embodiment of the application is shown in fig. 2, and the flowchart shown in fig. 2 is described as follows.
Step 201: the method comprises the steps of obtaining at least two categories of contents to be recommended of a target user, wherein each category comprises at least one content to be recommended.
Before step 201, a recommendation server may receive a recommendation request initiated by a client, where the recommendation request may carry a user identifier, for example, an account identifier of an application account where a user logs in, and determine a target user through the account identifier, so as to determine that content needs to be recommended to the target user, for example, content such as recommendation information or recommendation short videos is recommended. The content may be classified, for example, the content may be classified according to classification attribute information of the content, and the information may be classified into categories such as science and technology category, entertainment category, sports category, military category, social category, financial category and the like according to a field to which the information belongs, so that, further, at least two categories of content corresponding to the target user may be obtained.
The correspondence with the target user may refer to that the contents to be recommended are matched with the target user, which may be implemented by an information filtering function of the recommendation system, for example, matching is performed according to user feature data of the target user, and the contents having a certain matching degree with the user feature data are recalled from a mass content pool maintained by the recommendation system, that is, the preference of the user may be determined by the user feature data, and then the contents matched with the preference of the user are selected, the selected contents may be stored in a list to be recommended, and each content to be recommended in the list to be recommended has a recommendation sequence, and the recommendation sequence of each content to be recommended is related to the matching degree between the recommendation sequence and the user feature data, specifically, the higher score value of the score is determined according to the matching degree between the content to the user feature data, the higher the probability that the user likes and clicks is, the more forward the corresponding recommendation sequence is.
Wherein the user characteristic data may comprise at least one of user attribute data and user operation data. The user attribute data is data reflecting the basic attributes of the user, and may include data such as age, gender, geographic location, academic calendar, work, preference labels, and the like. The user operation data is data related to the operation of the user on the recommended content at ordinary times, for example, data related to the operation such as browsing duration, browsing times, whether to comment, whether to approve, whether to forward, whether to download, and the like on the recommended content. The interest and the preference of the user to the recommended content can be reflected to a certain extent through the user attribute data and the user operation data, so that the recommended content can be the content which is interested by the user with high probability through recalling the content to be recommended from the content library through the user characteristic data and sequencing the recommendation sequence according to the matching degree, and the content recommendation accuracy is improved.
In this embodiment of the application, the at least two categories of contents to be recommended acquired in step 201 may be contents selected from the list to be recommended recalled and sorted according to the user feature data, and if the contents to be recommended are subsequently sorted again, the contents to be recommended are equivalently sorted in a diversity manner again, so that the diversity rearrangement recommendation is performed on the basis of ensuring the recommendation accuracy, so as to improve the recommendation effectiveness as much as possible. Alternatively, the aforementioned at least two categories of content to be recommended may also be obtained in other manners, and the embodiment of the present application is not limited.
When selecting at least two categories of contents to be recommended from the list to be recommended, the total number of the contents to be recommended in the list to be recommended may be determined, and further, the size of the contents to be recommended may be compared with a set threshold (e.g., a first threshold: 150) according to the total number, and at least two categories of contents to be recommended for re-ranking may be selected from the list to be recommended according to a comparison result.
When the total number is greater than or equal to the first threshold, part of the plurality of contents to be recommended belonging to the plurality of categories can be selected, and equivalently, a part of the contents to be recommended is subjected to subsequent diversity reordering, so that the calculation amount caused by reordering in the recommendation process can be reduced, the recommendation delay and pause are avoided as much as possible, and the recommendation efficiency is improved.
When the total number is smaller than the first threshold, all the contents to be recommended in the list to be recommended may be selected as the contents to be reordered. It can be understood that if the content to be recommended in the list to be recommended at this time is less than two categories, some other types of content can be recalled from the content pool according to the user characteristic data to complete the content to be recommended of at least two categories, so as to meet the requirements for multiple categories; or, some other categories of associated content may be selected from the to-be-recommended list of the user associated with the target user (e.g., a social friend of the target user or other family members) to perform at least two categories of completion, because the target user is generally the same as or close to the life circle level of the user associated with the target user, and there may also be a certain similarity between the interested and concerned content, so that the content obtained from the associated user is likely to be of interest to the user himself, which is also in order to improve the accuracy of recommendation.
When selecting parts of the to-be-recommended contents belonging to at least two categories from the to-be-recommended list, the to-be-recommended contents with high matching degree may be selected according to the matching degree of each of the to-be-recommended contents in the to-be-recommended list, for example, some contents are selected for each category from the to-be-recommended contents with high matching degree to participate in subsequent diversity rearrangement, or multiple types of contents with matching degrees larger than a set threshold (for example, a second threshold) are selected to participate in subsequent diversity rearrangement, or the to-be-recommended contents with continuous ranking and belonging to at least multiple categories are selected from the to-be-recommended contents with highest matching degree in the to-be-recommended list to participate in subsequent diversity rearrangement, and so on. In the selection process, the number can also be limited, for example, only 100 contents to be recommended need to be selected to participate in the subsequent diversity rearrangement, so that the comprehensive selection can be performed by combining the aforementioned matching degree and the number requirement here.
Taking recommendation information as an example, it is assumed that 100 contents to be recommended are obtained, and the 100 contents to be recommended respectively belong to four types, i.e., science and technology, entertainment, sports, and military, and the 100 contents to be recommended already have a recommendation sequence, for example, the existing recommendation sequence is called as an original sequence. Referring to fig. 3, in the original ranking, it is assumed that the recommendation sequence is gradually backward from content 1, content 2, content 3, … …, and content 100, and the backward recommendation sequence is indicated in the direction indicated by the arrow in fig. 3.
Step 202: performing multiple rounds of selection on at least two categories of contents to be recommended to determine the contents selected in each round; wherein, in each round of selection: and respectively determining a recommended contribution predicted value of each candidate content in the current round relative to the category set to which the candidate content belongs, determining at least one candidate content of which the recommended contribution predicted value meets a set condition from all the candidate contents in the current round as the selected content in the current round, removing the selected content from the candidate contents in the current round, and correspondingly adding the selected content into the category set to which the candidate content belongs.
Step 203: and determining the recommendation sequence of the contents selected in each round according to the selection rounds.
In each round of selection, the contents to be recommended participating in the selection of the current round are called candidate contents, and the contents to be recommended selected in the current round and needing to be sorted in the current round are called selected contents in the current round.
The recommendation contribution prediction value of the content to be recommended is a quantitative value of the predicted recommendation benefit of the recommended content after being recommended, the larger the recommendation contribution prediction value is, the larger the probability that the content to be recommended is clicked by a user after being recommended is represented, that is, the higher the recommendation benefit generated by the content to be recommended after being recommended is represented, that is, the greater the contribution to the whole recommendation effect is, the recommendation benefit of the content to be recommended can be represented as the recommendation effect generated by the content to be recommended after being recommended, the better the recommendation effect is, and the greater the corresponding recommendation benefit is. Taking the content to be recommended as information as an example, the recommendation effect can be represented by the click rate of the user and the operation degree (such as the click viewing times, the browsing duration, whether to approve, whether to forward or comment, and the like) of the content to be recommended.
The recommendation contribution prediction value of each candidate content relative to the category set to which the candidate content belongs can be calculated according to the following modes: the method includes the steps of firstly calculating a total recommended contribution predicted value of all selected contents in a category set to which the candidate content belongs, for example, recording the total recommended contribution predicted value as a first total recommended contribution predicted value, then calculating a total recommended contribution predicted value of the candidate content and all selected contents in the category set to which the candidate content belongs, for example, recording the total recommended contribution predicted value as a second total recommended contribution predicted value, and finally determining a difference value between the second total recommended contribution value and the first total recommended contribution value corresponding to the candidate content as a recommended contribution predicted value of the candidate content, that is, determining a recommended contribution prediction increment of the candidate content relative to all selected contents in the category set to which the candidate content belongs as the recommended contribution predicted value of the candidate content.
In the embodiment of the application, the principle of marginal benefit decrement is utilized to realize the reordering of the diversity of the contents to be recommended, and the submodular function is suitable for processing the tasks of marginal benefit decrement, so that the recommendation contribution prediction value of each candidate content can be calculated through the submodular function.
The second order modulo function is defined as follows:
for a set S of items of a certain type, a subset thereof
Figure BDA0002582362750000151
Article x, y ∈ S:
f (A ≦ x } f (A ≦ x) f (A ≦ y) f (A ≦ x, y }) -f (A ≦ y }) (formula 1)
Functions that conform to the definition of equation 1 may be considered to be submodular functions, such as:
log(1+∑iscore(ai) (formula 2)
In equation 2, A is a set of classes CiA subset ofiIs an element in A, such as each science and technology information in the science and technology information set, aiThe yield of (c) is denoted score (a)i) For example, the profit is usually the estimated click rate, the estimated conversion rate, etc.
For the selected multiple categories of contents to be recommended, assuming that the number of categories is C, for example, the information may be classified into categories such as military category, entertainment category, scientific category, sports category, and the like. The C class sets each have a submodel function corresponding to each class set
Figure BDA0002582362750000152
The secondary modular yield of a is represented by the following equation 3:
Figure BDA0002582362750000153
the secondary modular total yield of these C sets is as follows:
Figure BDA0002582362750000161
the problem turns into:
Figure BDA0002582362750000162
the sub-optimal, i.e. near-optimal, solution of the submodular function may be solved by, for example, a greedy algorithm or other algorithm. This near optimal solution indicates that the total profit of the top 5 contents (top 5) combinations reordered by taking the submodular function can be guaranteed to be the maximum by the near optimal solution, which means that, for example, 5 contents need to be returned from 100 contents, there are many combinations, and each combination has a total profit.
For ease of understanding, the following examples are given.
As shown in fig. 3, it is assumed that the at least two categories of contents to be recommended acquired in step 201 that need to be subjected to diversity re-ranking are content 1, content 2, content 3, … …, and content 100, and the original ranks of the respective contents to be recommended gradually get back according to the order from small to large, and the 100 contents belong to four categories, namely military category, entertainment category, scientific category, and sports category.
Then, the 100 pieces of content are selected in multiple rounds to determine the selected content in each round of selection, that is, the selected content in each round of selection is obtained, and then the selected content in each round is sorted according to the selection rounds.
The first round of selection:
traversing the 100 contents, the recommended contribution prediction value of each candidate content can be calculated by using a submodular function, i.e. the submodular gain of each candidate content can be calculated.
Before the first selection round is performed on 100 contents, each category set does not include contents, that is, each category set is empty, and as shown in fig. 3, a rectangular frame representing 4 category sets does not include any contents, in the first selection round, each category set does not currently include any selected contents, so that the current total recommendation contribution prediction value of each category set is 0, that is, the current secondary modular revenue of each category set is 0.
Assuming that content 1 is a "military" classification and the current secondary modular revenue for the set of military classifications is 0, if content 1 is added to the set of military classifications and the secondary modular revenue for the entire set of military classifications is a1, the incremental revenue for content 1 is a1-0, i.e., a 1. It should be noted that content 1 is not actually added to the military classification set at this point, so it is more certain that a1 is the predicted revenue increment.
Similarly, content 2 has a revenue increment of b1, content 3 has a revenue increment of c1, … …, and content 100 has a revenue increment of z 1.
Comparing the profit gains of the candidate contents from the content 1 to the content 100, selecting the candidate content with the profit gains satisfying the setting condition as the selected content of the current round, wherein the setting condition is, for example, selecting the candidate content with the largest profit gain as the selected content of the current round, and assuming that the candidate content with the largest profit gain in the first round is the content 3, the classification is "science and technology". After the first round, as shown in fig. 4, the selected content in the first round selected is content 3, and at the same time, the content 3 is put into the 1 st position in the final recommendation list, as indicated by the dashed arrow for "content 3" in fig. 4, that is, the selected content in the first round is ranked in the first place of reordering, because each category set in the first round is empty, the candidate content with the largest profit increment indicates the largest contribution to the category to which it belongs, and the contribution to the overall profit of all the contents to be recommended is the largest among the candidate contents in all categories, so the reordered recommendation order thereof can be ranked in the first place.
Meanwhile, the content 3 is removed from the candidate content in the first round and put into the category set to which the content belongs, i.e. into the category set of "science and technology", as indicated by the solid arrow for "content 3" in fig. 4, so that after the first round, the "science and technology" category set includes the selected content of content 3, and the remaining candidate content does not include content 3 any more, i.e. content 3 is removed from the original ordering.
The purpose of removing content 3 from the candidate content is to not take it into account for the next round of selection, i.e. to not participate in the selection as a candidate content for the next round, since content 3 has been variously re-ordered in the first round as having been determined to be a selected content for the first round. In this regard, the candidate content in each selection round is not determined as the selected content before the selection round, and the candidate content in each selection round is not selected before the selection round because the candidate content determined as the selected content before the selection round has been added to the category set to which the candidate content belongs in the selection round determined as the selected content, such as the content 3 in fig. 4 has been added to the "technology class" to which the candidate content belongs in the first selection round.
And a second round of selection:
the only 99 candidates involved in the second round of selection to calculate the sub-modulo gain are because content 3 has been removed from the candidates in the previous round and added as selected to the category set of "technology classes". For the remaining 99 candidates, the submode gains for each candidate are calculated sequentially using the method described above.
Content 1 is a "military" classification, the set of classes for which the military classification is currently still empty, so the secondary modulus gain for content 1 is still a 1.
Content 2 is the "entertainment" category and the revenue increment for content 2 is b 1.
Content 4 is the "sports" category and the revenue increment for content 4 is d 1.
Content 5 is the "science and technology" category, and since there is currently selected content 3 in the category set of the "science and technology" category, the current submode gain of the category set is not 0, and f represents the submode function, then in the second round, the gain e2 of content 5 is f ({ content 5, content 3}) -f ({ content 3 }).
……
Content 17 is the "entertainment" category and the revenue increment for content 17 is, for example, x 1.
Similarly, the gain increments for all other candidate content in the second round may be calculated.
After the gain increments of the candidate contents are obtained in the second round, all the gain increments may be compared, and the candidate contents whose gain increments satisfy the setting conditions are selected as the selected contents in the second round.
In one possible implementation, the candidate content with the largest profit increment may be taken as the selected content of the current round, and assuming that the profit increment x1 of the content 17 is the largest in the second round, the content 17 may be taken as the selected content of the current round and added to the second recommendation position in the final recommendation list and removed from the candidate content of the current round while the content 17 is added to the category set of its corresponding "sports class", as shown in fig. 5 a. In this way, the candidate content with the largest profit increment in the current round is selected as the selected content in the current round and is reordered preferentially, so that the overall profit of the obtained final recommendation list can be ensured to be maximized as much as possible, and the recommendation effectiveness is further improved.
In another possible implementation, a predetermined number of candidate contents from the maximum recommended contribution prediction value may also be determined in the descending order of the recommended contribution prediction value, where the predetermined number is an integer greater than or equal to 2, for example, 2, or 3, or 4. Further, the selected content of the current round is determined from the selected predetermined number of candidate content. Assuming that the predetermined number is 2, the 2 candidate contents with the largest profit increments selected in the order of the profit increments from large to small are the content 17 and the content 5, i.e., the content 5 is the candidate content with the second largest profit increment in the second selection.
Further, in a case where the gain increment between a predetermined number of candidate contents is not very different, for example, when the difference between the recommendation contribution prediction values of two adjacent candidate contents in the predetermined number of candidate contents is smaller than a difference threshold (for example, a third threshold), the predetermined number of candidate contents may all be determined as the selected contents of the current round, for example, the difference between the submodular increments of the content 17 and the content 5 is smaller than the aforementioned third threshold, as shown in fig. 5b, the content 17 and the content 5 may be simultaneously selected as the selected contents of the current round, as indicated by dotted arrows for "content 5" and "content 17" in fig. 5b, and are sequentially ranked in the final recommendation list according to the magnitude of the gain increment, that is, the ranking position in the final recommendation list is the farther the larger the gain increment is. By comparing the difference between the recommendation contribution prediction increments of two adjacent candidate contents with the difference threshold, it can be ensured that the submode gains between the selected predetermined number of candidate contents are not too different, that is, the contribution of each candidate content in the predetermined number of candidate contents to the final recommendation total profit is not too much, so that the candidate contents can be subjected to diversity rearrangement.
In yet another embodiment, when the difference between the recommendation contribution prediction values of two adjacent candidate contents in the predetermined number of candidate contents is smaller than the difference threshold (for example, the aforementioned third threshold), the candidate contents meeting the prioritization condition may be further selected from the predetermined number of candidate contents as the selected contents of the current round. Therefore, the candidate contents meeting the priority ordering condition can be prioritized and ordered, namely, the candidate contents can be arranged at the front position, so that the maximum recommendation total income can be ensured as much as possible, and the effectiveness and the accuracy of recommendation are improved.
For example, a candidate content different from the category of the selected content selected in the previous round among the predetermined number of candidate contents is used as the selected content in the current round, the selected content selected in the previous round, i.e., the first round, is the content 3 and is the technology category, and the content 17 and the content 5 are not the technology category, so that the current round can determine the content 17 and the content 5 as the selected content at the same time and prioritize the selected content, as shown in fig. 5 b. Assuming again that the content 17 is a science and technology class and the content 5 is not a science and technology class and the content 5 is an entertainment class, in this manner, only the content 5 may be selected as the selected content, as shown in fig. 5 c. Therefore, the contents selected by two adjacent rounds are different types of contents, and the different types of contents can be arranged in a cross mode as much as possible, so that the diversity is enhanced.
Or, for example, a target category whose number of selected contents included in at least two categories before the current round of selection meets a set number limit is determined, and a candidate content belonging to the target category among a predetermined number of candidate contents is taken as the selected content of the current round. For example, there are four categories of "science and technology", "military", "entertainment", and "sports", if the selection is performed in the 10 th round, and the content selected as "science and technology" in all the previous 9 rounds is the least or less than the specified number, the "science and technology" may be determined as the target category in the embodiment of the present application, and further, the selection may be performed in this round with the content belonging to the "science and technology" as the selection result, so as to enhance the diversity as much as possible and ensure that the number of each type is not too small.
For another example, the candidate content with the highest original ranking of the predetermined number of candidate contents is used as the selected content of the current round, wherein the original ranking of the candidate contents refers to ranking in the order of high to low matching degree between the candidate contents and the user feature data of the target user. In the second round, the content 17 is earlier than the content 5 in the original ordering of the content 5, so as shown in fig. 5c, only the content 5 may be determined as the selected content of the current round and the ordering may be performed preferentially, so that in the case that the modulo gains of the two candidate contents are not much different, in order to simultaneously consider the accuracy of the recommendation, the priority of the earlier ordering may be re-ordered, so that the requirement of the accuracy of the recommendation may be satisfied as much as possible on the basis of satisfying the diversity re-ordering.
Referring to the above-described manner of selecting the selected content in the first round and the second round, multiple rounds of selection and reordering may be performed until all the contents to be recommended are selected as the selected content and reordered. For 100 contents to be recommended, assuming that only one candidate content is selected in each round as the selected content of the current round, 100 rounds of selection are required, if a plurality of contents are selected in some rounds as the selected content of the current round, less than 100 rounds of selection are required until all the candidate contents in the 100 candidate contents are sorted into the final recommendation list, and thus a new rearrangement sequence list is formed.
Step 204: and recommending the contents to the target user according to the sequence of the contents to be recommended which are arranged in the recommendation sequence.
Further, after the final recommendation list with the re-ordered diversity is obtained, the content recommendation can be sequentially performed to the user according to the recommendation order in the final recommendation list.
It should be noted that, if the contents to be recommended in at least two categories in step 201 are obtained by selecting from the to-be-recommended list obtained according to the precision recommendation, the obtained final recommendation list may be spliced with the remaining contents to be recommended in the to-be-recommended list, and then recommendation is sequentially performed according to the spliced recommendation order.
In the embodiment of the present application, reordering of multiple categories of content is implemented by using marginal benefit decrement, please refer to a schematic diagram of marginal benefit decrement shown in fig. 6, for a same category of articles, the total recommended revenue is gradually increased along with the increase of the number of recommended articles, but the increase of the total recommended revenue is gradually decreased along with the gradual increase of the number of recommended articles, so that the total revenue curve in fig. 6 is gradually gentle along with the increase of the number of recommended articles, for this reason, the increase of the total recommended revenue is improved by increasing the categories of recommended content, so as to improve the total recommended revenue as much as possible.
Therefore, in the embodiment of the application, the final recommendation list is reordered according to the recommendation contribution predicted values of the contents to be recommended relative to the category set to which the contents to be recommended belong, so that the probability that the contents to be recommended before being reordered belong to a plurality of categories is high, the contents to be recommended in the categories are recommended to the user according to the order after being reordered, and the contents to be recommended in the categories can be recommended to the user with high probability, so that the diversity of the recommended contents can be improved, and the recommendation redundancy is reduced to a certain extent.
In order to facilitate understanding of the multiple reordering schemes in the embodiments of the present application, the technical solution in the embodiments of the present application is further described below with reference to fig. 7 by taking the common information recommendation as an example.
Step 701: and acquiring an information recommendation request of a target user.
Taking a target user as an example, when the target user wishes to view information, the information recommendation request may be triggered to the recommendation server through the client, for example, when the user opens an information application, the client may be directly triggered to send the information recommendation request to the recommendation server, or after the user may perform a specific operation in an application interface of the information application or input a request of "latest information", the client may be triggered to send the information recommendation request to the recommendation server.
Step 702: and recalling a plurality of pieces of information to be recommended from the information pool according to the user characteristic data of the target user.
The information pool is a database which is maintained by the recommendation system and comprises mass information, information issued by a user through the recommendation system can be stored in the information pool, and in order to improve storage performance, the information pool can be cleaned regularly, for example, information issued half a year ago is deleted from the information pool, so that more newly issued information can be stored.
According to the user characteristic data of the target user, some information which is matched with the user characteristic data can be selected from the information pool as recall information, the matching degree of the information and the user characteristic data can represent the user's liking degree on the information, for example, the information with higher matching degree is recommended to the target user, and the probability that the target user clicks the information is higher, so that more accurate recommendation can be realized.
Generally, when information to be recommended is recalled from an information pool, the amount of recalls is not too small, so that a certain amount of information can be continuously recommended to a target user, and although the recall is performed according to the user characteristic data of the target user, the recall amount is large and the user characteristic data can be matched with various types of information, so that the recalled information to be recommended generally belongs to various types.
And information is recalled from the information pool according to the user characteristic data, so that the recalled information to be recommended is ensured to be the information which is liked by the target user with high probability, and the recommendation accuracy can be improved as much as possible.
Step 703: and according to the matching degree of each piece of information to be recommended and the user characteristic data, scoring the matching degree of the recalled contents to be recommended.
For example, the matching degree between each piece of information to be recommended and the user feature data can be predicted through a matching degree prediction model, the matching degree of each piece of recalled information to be recommended is scored in a mode that the higher the matching degree is, the higher the corresponding score is, for example, the scoring range is 50-100 scores, the higher the score is, the greater the likeness degree of the user to the information is, and thus, the greater the probability that the target user clicks the information is after the higher the score is recommended to the target user.
Step 704: and sequencing according to the sequence of the matching degree from high to low to obtain a sequenced list to be recommended.
If the number of the information to be recommended recalled from the information pool is 150, the 150 pieces of information can be sorted according to the high-low matching degree score, the higher the matching degree score is, the higher the information is, so as to obtain a sorted list to be recommended, it can be understood that the first information sorted in the sorted list to be recommended is the information with the highest matching degree score, and the lower the matching degree score of the corresponding information is from the first information sorted in sequence.
Step 705: and selecting K contents to be recommended from the list to be recommended, which are ranked in the front and belong to a plurality of categories, and recording the K contents to be recommended as a topK list.
In the sorted to-be-recommended list, information to be reordered can be selected from the first sorted information, taking the value of K as 100 as an example, the information of 100 before sorting can be selected from the 150 pieces of information already sorted in the front as a topK list, it needs to be noted that the information to be recommended included in the topK list belongs to at least two categories, if the categories to which the information to be recommended of 100 before sorting belongs are less than two, the information to be recommended of 100 after sorting can be selected to be sorted as far as possible to fill up the at least two categories, so that the list to be recommended in the selected topK list is matched with the target user as far as possible and belongs to multiple categories, and thus, the subsequent reordering can be performed on the basis of ensuring the accuracy, so that the accuracy and the diversity of the recommended content can be simultaneously ensured.
Step 706: the topK list is traversed.
After the topK list is obtained, multiple rounds of selection can be performed on all the information to be recommended included in the topK list according to the recommendation contribution prediction values of the information to be recommended relative to the category set to which the information belongs.
Step 707: and judging whether the recommended contribution predicted value of the currently traversed information is maximum or not.
For the topK list, traversing each piece of information to be recommended according to a certain sequence, sequentially calculating a recommendation contribution prediction value of each piece of information to be recommended relative to the category set to which the information to be recommended belongs, and comparing the recommendation contribution prediction value of the currently traversed information to be recommended relative to the category set to which the information to be recommended belongs with the recommendation contribution prediction value of the previously traversed information to be recommended to judge whether the recommendation contribution prediction value of the currently traversed information is the maximum, that is, in the traversing process, the maximum recommendation contribution prediction value can be determined from all pieces of information to be recommended.
Step 708: and when the maximum recommended contribution predicted value of the currently traversed information is determined, judging whether the topK list is traversed or not.
Step 709: and judging whether the recommended contribution predicted value of the next information is maximum or not.
When the maximum recommended contribution prediction value of the currently traversed information is determined, whether the topK list is traversed or not is judged to find out the maximum recommended contribution prediction value from all recommended contribution prediction values corresponding to all information to be recommended. That is, the information to be recommended with the largest recommendation contribution prediction value needs to be selected from all the categories of information to be recommended, because the information to be recommended with the largest recommendation contribution prediction value indicates that the recommendation profit is the largest for all the information to be recommended in the topK list, so that it can be selected as the first ranking in the diversity ranking.
When the fact that the recommended contribution predicted value of the currently traversed information is not the largest of all the obtained recommended contribution predicted values is determined, or when the fact that the recommended contribution predicted value of the currently traversed information is the largest of all the obtained recommended contribution predicted values and a topK list is not traversed yet is determined, the recommended contribution predicted value of the next information can be continuously calculated according to the traversal sequence, and whether the recommended contribution predicted value of the next information is the largest of all the obtained recommended contribution predicted values is judged.
Step 710: and putting the information with the maximum recommendation contribution predicted value into a diversity list for priority ranking.
That is to say, according to the traversal order, the recommendation contribution prediction values of the contents to be recommended in the topK list are sequentially calculated, the information with the maximum recommendation contribution prediction value is selected from the topK list, and the selected information with the maximum recommendation contribution prediction value is put into the diversity list to be preferentially sorted, for example, the information with the maximum recommendation contribution prediction value selected in the first round is arranged in the first sorting of the diversity list.
Step 711: and removing the information with the maximum recommendation contribution predicted value from the topK list and adding the information into the category set to which the information belongs.
After sorting the recommendation contribution predictors in the diversity list, it may be removed from the topK list, after which substantially only K-1 pieces of information to be recommended are in the topK list, i.e. the topK list is actually updated.
Step 712: it is determined whether the topK list is empty.
By adopting the manner of the above steps 706 to 711, the updated topK list is selected for multiple rounds again until all the information to be recommended in the topK list is removed and added to the category set corresponding to each piece of information to be recommended, at this time, all the information to be recommended in the original topK list has new ordering in the diversity list, and finally, the topK list becomes an empty list.
Step 713: and when the topK list is empty, replacing the topK list selected from the list to be recommended with the diversity list to obtain a reordered list to be recommended.
Step 714: and recommending information to the target user according to the recommendation sequence in the reordered list to be recommended.
After the topK list is changed into an empty list, the diversity list carries out diversity reordering on 100 pieces of information to be recommended in the original topK list, and furthermore, the obtained diversity list can replace all information to be recommended at K before the ranking (namely 100 before the ranking) in the original ranked list to be recommended, so that the ranking of the information at the top 100 after reordering is not the same as the ranking of the information at the top 100 in the original list to be recommended, compared with the information at the top 100 in the original list to be recommended, the information of various types in the information at the top 100 after reordering is scattered as much as possible, for example, the information of various types is scattered and arranged in a staggered and crossed way, thus when recommending information to a target user, the information of various types can be preferentially recommended to the target user as much as possible, thereby improving the diversity of the types of the recommended information, and simultaneously, the recommending sequence can be diversified as much as possible, the recommendation redundancy caused by recommending single-type information is reduced as much as possible, and the recommendation effectiveness is improved.
The embodiment of the application is a technical scheme applied to a recommendation scene to solve the diversity recommendation problem. The diversity recommendation problem in the recommendation scene is a problem of how to recommend different types of recommended content according to a reasonable sequence. It is understood that when different recommended contents are recommended in different orders, a user may generate different browsing behavior data when browsing the recommended contents. For example, under a recommendation mechanism only considering accuracy in the related art, the recommendation system recommends 8 pieces of entertainment information and 1 piece of sports information to the user, and the user clicks 3 pieces of entertainment information, because the entertainment information is too much and the content may be repeated, the overall click rate of the user on the recommended 9 pieces of information is not high. Supposing that by adopting the recommendation mechanism after the diversity reordering provided by the embodiment of the present application, the recommendation system also recommends 9 pieces of information to the user, wherein there are 5 pieces of entertainment information, 2 pieces of sports information, 1 piece of science and technology information, and 1 piece of military information, and the user clicks 2 pieces of entertainment information, 1 piece of sports information, 1 piece of science and technology information, and 1 piece of military information, it can be seen that the overall click rate of the user on the 9 pieces of information is much higher, which is equivalent to improving the overall profit of information recommendation, by providing diverse information recommendations for the user, not only the overall profit of recommendation can be improved, thereby improving the accuracy and effectiveness of recommendation, but also guiding the user to find some potential interests as much as possible, and these diverse content recommendations may bring unexpected surprise and surprise to the user, thereby enhancing the user experience, the user viscosity is improved, and the user retention rate is improved.
In the embodiment of the application, the contents to be recommended are subjected to multi-class reordering by utilizing the principle of diminishing marginal benefits, so that diversified recommendation ordering is realized, after the contents to be recommended are reordered, the contents of multiple classes are scattered as much as possible and are arranged in a cross mode, in the process of sequentially recommending according to the re-ordered recommendation sequence, the contents of multiple classes can be recommended, the purpose of improving the diversity of the recommended contents is realized, surprising and recommendation experience can be provided for a user, the potential favor of the user can be mined to a certain extent, the viscosity of the user is further improved, and the retention rate of the user is improved. Moreover, the recommendation contribution value of the content at the front is relatively larger for all the contents to be recommended, so that the sequential recommendation can also ensure that the overall profit of all the recommended contents to be recommended is maximized as much as possible, thereby ensuring the accuracy and effectiveness of the recommendation.
Based on the same inventive concept, the embodiment of the present application provides a device for recommending content, which may be a hardware structure, a software module, or a hardware structure plus a software module. The content recommending device is, for example, the server 104 in fig. 1, or may be a functional device disposed in the server 104, and the content recommending device may be implemented by a chip system, and the chip system may be formed by a chip, and may also include a chip and other discrete devices. Referring to fig. 8, the apparatus for recommending content in the embodiment of the present application includes an obtaining module 801, an ordering module 802, and a recommending module 803, where:
an obtaining module 801, configured to obtain at least two categories of to-be-recommended content of a target user, where each category includes at least one to-be-recommended content;
the sorting module 802 is configured to perform multiple rounds of selection on the contents to be recommended of at least two categories according to the recommendation contribution prediction values, and determine a recommendation order of each round of selected contents according to selection rounds until all the contents to be recommended are sorted; wherein, in each round of selection: respectively determining a recommended contribution predicted value of each candidate content in the current round relative to the category set to which the candidate content belongs, determining at least one candidate content of which the recommended contribution predicted value meets a set condition from all the candidate contents in the current round as the content selected in the current round, and removing the selected content from the candidate content and correspondingly adding the selected content into the category set to which the candidate content belongs; wherein the candidate content in each round of selection is not selected before the round of selection;
and the recommending module 803 is configured to recommend the content to the target user according to the sequence of the contents to be recommended arranged in the finished recommending sequence.
In one possible implementation, the obtaining module 801 is configured to:
acquiring a to-be-recommended list corresponding to a target user, wherein the to-be-recommended content in the to-be-recommended list is recalled according to user characteristic data of the target user, and the recommendation sequence of the to-be-recommended content in the to-be-recommended list is related to the matching degree between each to-be-recommended content and the user characteristic data;
and selecting at least two categories of contents to be recommended from the contents to be recommended with high matching degree according to the matching degree of each content to be recommended in the list to be recommended.
In one possible implementation, the ordering module 802 is configured to:
and selecting the candidate content with the largest recommendation contribution predicted value from all the candidate contents in the current round as the selected content in the current round.
In one possible implementation, the ordering module 802 is configured to:
determining a preset number of candidate contents from the maximum recommended contribution predicted value according to the descending order of the recommended contribution predicted value, wherein the preset number is an integer greater than or equal to 2;
the content selected in the current round is determined from a predetermined number of candidate content.
In one possible implementation, the ordering module 802 is configured to:
and according to the sequence of the recommended contribution predicted values from large to small, if the difference value between the recommended contribution predicted values of every two adjacent candidate contents in the preset number of candidate contents is smaller than the difference threshold value, determining the selected contents in the current round from the preset number of candidate contents.
In one possible implementation, the ordering module 802 is configured to:
determining all the candidate contents in a preset number as the contents selected in the current round; alternatively, the first and second electrodes may be,
and selecting candidate contents meeting the prioritization condition from the preset number of candidate contents as the contents selected in the current round.
In one possible implementation, the ordering module 802 is configured to:
taking candidate contents with different categories from the contents selected in the previous round in the preset number of candidate contents as the contents selected in the current round; alternatively, the first and second electrodes may be,
and determining the target category of which the number of the selected contents included in at least two categories before the current round of selection meets the set number limit, and taking the candidate contents belonging to the target category in the preset number of candidate contents as the selected contents in the current round.
In one possible implementation, the ordering module 802 is configured to:
and taking the candidate content with the highest original ranking of the candidate contents in a preset number as the content selected in the current round, wherein the original ranking of the candidate contents is ranking according to the sequence from high to low of the matching degree between the candidate contents and the user characteristic data of the target user.
In one possible implementation, the ordering module 802 is configured to:
calculating a first total recommendation contribution prediction value of all selected contents in the category set to which each candidate content belongs through a submodular function;
calculating a second total recommendation contribution prediction value of each candidate content and all selected contents in the category set to which the candidate content belongs through a secondary modulus function;
and determining the difference value of the second total recommended contribution predicted value of each candidate content relative to the first total recommended contribution predicted value as the recommended contribution predicted value of the candidate content.
In one possible implementation, the ordering module 802 is configured to:
the recommendation order is arranged for the contents selected in each round in such a way that the recommendation order of the contents corresponding to the selection is earlier in the selection round.
All relevant contents of each step involved in the foregoing embodiments of the method for recommending contents may be cited to the functional description of the functional module corresponding to the apparatus for recommending contents in the embodiments of the present application, and are not described herein again.
The division of the modules in the embodiments of the present application is schematic, and only one logical function division is provided, and in actual implementation, there may be another division manner, and in addition, each functional module in each embodiment of the present application may be integrated in one processor, may also exist alone physically, or may also be integrated in one module by two or more modules. The integrated module can be realized in a hardware mode, and can also be realized in a software functional module mode.
Based on the same inventive concept, an embodiment of the present application provides a computing device, for example, the aforementioned server 104 in fig. 1, the computing device is capable of executing the method for recommending content provided by the embodiment of the present application, as shown in fig. 9, the computing device in the embodiment of the present application includes at least one processor 901, and a memory 902 and a communication interface 903 connected to the at least one processor 901, a specific connection medium between the processor 901 and the memory 902 is not limited in the embodiment of the present application, in fig. 9, the processor 901 and the memory 902 are connected through a bus 900 as an example, the bus 900 is represented by a thick line in fig. 9, and a connection manner between other components is merely illustrated schematically and is not limited. The bus 900 may be divided into an address bus, a data bus, a control bus, etc., and is shown with only one thick line in fig. 9 for ease of illustration, but does not represent only one bus or type of bus.
In the embodiment of the present application, the memory 902 stores a computer program executable by the at least one processor 901, and the at least one processor 901 may execute the steps included in the foregoing method for recommending content by executing the computer program stored in the memory 902.
The processor 901 is a control center of the computing device, and may connect various parts of the entire computing device by using various interfaces and lines, and perform various functions and process data of the computing device by executing or executing instructions stored in the memory 902 and calling data stored in the memory 902, thereby performing overall monitoring on the computing device. Optionally, the processor 901 may include one or more processing modules, and the processor 901 may integrate an application processor and a modem processor, where the processor 901 mainly processes an operating system, a user interface, an application program, and the like, and the modem processor mainly processes wireless communication. It will be appreciated that the modem processor described above may not be integrated into the processor 901. In some embodiments, the processor 901 and the memory 902 may be implemented on the same chip, or in some embodiments, they may be implemented separately on separate chips.
The processor 901 may be a general-purpose processor, such as a Central Processing Unit (CPU), a digital signal processor, an application specific integrated circuit, a field programmable gate array or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof, that may implement or perform the methods, steps, and logic blocks disclosed in the embodiments of the present application. A general purpose processor may be a microprocessor or any conventional processor or the like. The steps of a method disclosed in connection with the embodiments of the present application may be directly implemented by a hardware processor, or may be implemented by a combination of hardware and software modules in a processor.
Memory 902, which is a non-volatile computer-readable storage medium, may be used to store non-volatile software programs, non-volatile computer-executable programs, and modules. The Memory 902 may include at least one type of storage medium, and may include, for example, a flash Memory, a hard disk, a multimedia card, a card-type Memory, a Random Access Memory (RAM), a Static Random Access Memory (SRAM), a Programmable Read Only Memory (PROM), a Read Only Memory (ROM), a charge Erasable Programmable Read Only Memory (EEPROM), a magnetic Memory, a magnetic disk, an optical disk, and so on. The memory 902 is any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer, but is not limited to such. The memory 902 of the embodiments of the present application may also be circuitry or any other device capable of performing a storage function for storing program instructions and/or data.
The communication interface 903 is a transmission interface capable of performing communication, and may receive data or transmit data through the communication interface 903, for example, data interaction may be performed with other devices through the communication interface 903 to achieve the purpose of communication.
Further, the computing device includes a basic input/output system (I/O system) 904 for facilitating information transfer between various devices within the computing device, and a mass storage device 908 for storing an operating system 905, application programs 906, and other program modules 907.
The basic input/output system 904 includes a display 909 for displaying information and an input device 910 such as a mouse, keyboard, etc. for user input of information. Wherein a display 909 and an input device 910 are connected to the processor 901 via a basic input/output system 904 connected to the system bus 900. The basic input/output system 904 may also include an input/output controller for receiving and processing input from a number of other devices, such as a keyboard, mouse, or electronic stylus. Similarly, an input-output controller may also provide output to a display screen, a printer, or other type of output device.
The mass storage device 908 is connected to the processor 901 through a mass storage controller (not shown) connected to the system bus 900. The mass storage device 908 and its associated computer-readable media provide non-volatile storage for the server package. That is, the mass storage device 908 may include a computer-readable medium (not shown), such as a hard disk or CD-ROM drive.
According to various embodiments of the present application, the computing device package may also be operated by a remote computer connected to the network through a network, such as the Internet. That is, the computing device may be connected to the network 911 via the communication interface 903 coupled to the system bus 900, or may be connected to another type of network or remote computer system (not shown) using the communication interface 903.
Based on the same inventive concept, the present application also provides a storage medium, which may be a computer-readable storage medium, and the storage medium stores computer instructions, which when executed on a computer, cause the computer to perform the steps of the method for recommending content as described above.
Based on the same inventive concept, the embodiment of the present application further provides a chip system, which includes a processor and may further include a memory, and is configured to implement the steps of the method for recommending content as described above. The chip system may be formed by a chip, and may also include a chip and other discrete devices.
In some possible implementations, various aspects of the method for recommending content provided by the embodiments of the present application can also be implemented in the form of a program product including program code for causing a computer to perform the steps of the method for recommending content according to various exemplary implementations of the present application described above when the program product is run on the computer.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, optical storage, and the like) having computer-usable program code embodied therein.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present application without departing from the spirit and scope of the application. Thus, if such modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is intended to include such modifications and variations as well.

Claims (15)

1. A method for recommending content, the method comprising:
acquiring at least two categories of contents to be recommended of a target user, wherein each category comprises at least one content to be recommended;
performing multiple rounds of selection on the contents to be recommended of the at least two categories according to the recommendation contribution prediction value, and determining the recommendation sequence of the contents selected in each round according to the selection rounds until all the contents to be recommended are sequenced; wherein, in each round of selection: respectively determining a recommended contribution predicted value of each candidate content in the current round relative to the category set to which the candidate content belongs, determining at least one candidate content of which the recommended contribution predicted value meets a set condition from all the candidate contents in the current round as the content selected in the current round, and removing the selected content from the candidate content and correspondingly adding the selected content into the category set to which the candidate content belongs; wherein the candidate content in each round of selection is not selected before the round of selection;
and recommending the contents to the target user according to the sequence of the contents to be recommended which are arranged in the recommendation sequence.
2. The method of claim 1, wherein obtaining at least two categories of content to be recommended for a target user comprises:
acquiring a to-be-recommended list corresponding to the target user, wherein the to-be-recommended content in the to-be-recommended list is recalled according to user characteristic data of the target user, and the recommendation sequence of the to-be-recommended content in the to-be-recommended list is related to the matching degree between each to-be-recommended content and the user characteristic data;
and selecting at least two categories of contents to be recommended from the contents to be recommended with high matching degree according to the matching degree of each content to be recommended in the list to be recommended.
3. The method according to claim 1, wherein determining at least one candidate content whose recommended contribution prediction value satisfies a set condition as the content selected in the current round from all the candidate contents in the current round comprises:
and selecting the candidate content with the largest recommendation contribution predicted value from all the candidate contents in the current round as the selected content in the current round.
4. The method according to claim 1, wherein determining at least one candidate content whose recommended contribution prediction value satisfies a set condition as the content selected in the current round from all the candidate contents in the current round comprises:
determining a preset number of candidate contents from the maximum recommended contribution predicted value according to the descending order of the recommended contribution predicted value, wherein the preset number is an integer greater than or equal to 2;
determining the content selected in the current round from the predetermined number of candidate contents.
5. The method of claim 4, wherein determining the content for the current selection from the predetermined number of candidate content comprises:
and according to the sequence of the recommended contribution predicted values from large to small, if the difference value between the recommended contribution predicted values of every two adjacent candidate contents in the predetermined number of candidate contents is smaller than the difference threshold value, determining the selected contents in the current round from the predetermined number of candidate contents.
6. The method of claim 4, wherein determining the content for the current selection from the predetermined number of candidate content comprises:
determining all the candidate contents in the preset number as the contents selected in the current round; alternatively, the first and second electrodes may be,
and selecting candidate contents meeting the priority ordering condition from the preset number of candidate contents as the contents selected in the current round.
7. The method of claim 6, wherein selecting, as the content of the current round of selection, a candidate content satisfying a prioritization condition from the predetermined number of candidate contents comprises:
taking the candidate content with the category different from that of the content selected in the previous round in the preset number of candidate contents as the content selected in the current round; alternatively, the first and second electrodes may be,
and determining a target category of which the number of the selected contents included in the at least two categories before the current round of selection meets a set number limit, and taking the candidate contents belonging to the target category in the predetermined number of candidate contents as the selected contents in the current round.
8. The method of claim 6, wherein selecting, as the content of the current round of selection, a candidate content satisfying a prioritization condition from the predetermined number of candidate contents comprises:
and taking the candidate content with the highest original ranking of the candidate contents in the preset number as the content selected in the current round, wherein the original ranking of the candidate contents is ranking according to the sequence from high to low of the matching degree between the candidate contents and the user characteristic data of the target user.
9. The method according to any one of claims 1 to 8, wherein the step of respectively determining the recommendation contribution predicted value of each candidate content in the current round relative to the category set to which the candidate content belongs comprises the following steps:
calculating a first total recommendation contribution prediction value of all selected contents in the category set to which each candidate content belongs through a submodular function;
calculating a second total recommendation contribution prediction value of each candidate content and all selected contents in the category set to which the candidate content belongs through the secondary modulus function;
and determining the difference value of the second total recommended contribution predicted value of each candidate content relative to the first total recommended contribution predicted value as the recommended contribution predicted value of the candidate content.
10. The method of any one of claims 1-8, wherein determining the order of recommendation for each round of selected content based on the round of selection comprises:
the recommendation order is arranged for the contents selected in each round in such a way that the recommendation order of the contents corresponding to the selection is earlier in the selection round.
11. An apparatus for recommending content, the apparatus comprising:
the system comprises an acquisition module, a recommendation module and a recommendation module, wherein the acquisition module is used for acquiring at least two categories of contents to be recommended of a target user, and each category comprises at least one content to be recommended;
the sequencing module is used for carrying out multiple rounds of selection on the contents to be recommended of the at least two categories according to the recommendation contribution prediction value, and determining the recommendation sequence of each round of selected contents according to the selection rounds until all the contents to be recommended are sequenced; wherein, in each round of selection: respectively determining a recommended contribution predicted value of each candidate content in the current round relative to the category set to which the candidate content belongs, determining at least one candidate content of which the recommended contribution predicted value meets a set condition from all the candidate contents in the current round as the content selected in the current round, removing the selected content from the candidate content and correspondingly adding the selected content into the category set to which the selected content belongs, wherein the candidate content in each round of selection is not selected before the round of selection;
and the recommending module is used for recommending the content to the target user according to the sequence of the contents to be recommended which are arranged in the recommending sequence.
12. The apparatus of claim 11, wherein the acquisition module is to:
acquiring a to-be-recommended list corresponding to the target user, wherein the to-be-recommended content in the to-be-recommended list is recalled according to user characteristic data of the target user, and the recommendation sequence of the to-be-recommended content in the to-be-recommended list is related to the matching degree between each to-be-recommended content and the user characteristic data;
and selecting at least two categories of contents to be recommended from the contents to be recommended with high matching degree according to the matching degree of each content to be recommended in the list to be recommended.
13. The apparatus of claim 11, wherein the ordering module is to:
and selecting the candidate content with the largest recommendation contribution predicted value from all the candidate contents in the current round as the selected content in the current round.
14. A computing device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the steps comprised by the method according to any one of claims 1 to 10 when executing the computer program.
15. A storage medium storing computer-executable instructions for causing a computer to perform the steps comprising the method of any one of claims 1-10.
CN202010671209.6A 2020-07-13 2020-07-13 Method and device for recommending content, computing equipment and storage medium Pending CN111782957A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010671209.6A CN111782957A (en) 2020-07-13 2020-07-13 Method and device for recommending content, computing equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010671209.6A CN111782957A (en) 2020-07-13 2020-07-13 Method and device for recommending content, computing equipment and storage medium

Publications (1)

Publication Number Publication Date
CN111782957A true CN111782957A (en) 2020-10-16

Family

ID=72767657

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010671209.6A Pending CN111782957A (en) 2020-07-13 2020-07-13 Method and device for recommending content, computing equipment and storage medium

Country Status (1)

Country Link
CN (1) CN111782957A (en)

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112364244A (en) * 2020-11-16 2021-02-12 广州酷狗计算机科技有限公司 Multimedia data recommendation method, device, server and storage medium
CN112784159A (en) * 2021-01-21 2021-05-11 中国平安人寿保险股份有限公司 Content recommendation method and device, terminal equipment and computer readable storage medium
CN112882929A (en) * 2021-02-02 2021-06-01 网易（杭州）网络有限公司 Test method, test device, computer equipment and storage medium
CN112905879A (en) * 2021-01-29 2021-06-04 北京达佳互联信息技术有限公司 Recommendation method, recommendation device, server and storage medium
CN113468426A (en) * 2021-07-01 2021-10-01 北京明略软件系统有限公司 Information recommendation method and device, electronic equipment and readable storage medium
CN113515696A (en) * 2021-05-17 2021-10-19 上海众源网络有限公司 Recommendation method and device, electronic equipment and storage medium
CN113553509A (en) * 2021-07-29 2021-10-26 北京达佳互联信息技术有限公司 Content recommendation method and device, electronic equipment and storage medium
CN113781086A (en) * 2021-01-21 2021-12-10 北京沃东天骏信息技术有限公司 Article recommendation method, device, medium and electronic equipment
CN116911957A (en) * 2023-09-12 2023-10-20 深圳须弥云图空间科技有限公司 Heterogeneous content mixed recommendation method, device, system and storage medium

Cited By (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112364244A (en) * 2020-11-16 2021-02-12 广州酷狗计算机科技有限公司 Multimedia data recommendation method, device, server and storage medium
CN113781086A (en) * 2021-01-21 2021-12-10 北京沃东天骏信息技术有限公司 Article recommendation method, device, medium and electronic equipment
CN112784159A (en) * 2021-01-21 2021-05-11 中国平安人寿保险股份有限公司 Content recommendation method and device, terminal equipment and computer readable storage medium
CN112784159B (en) * 2021-01-21 2022-12-09 中国平安人寿保险股份有限公司 Content recommendation method and device, terminal equipment and computer readable storage medium
CN112905879B (en) * 2021-01-29 2024-03-19 北京达佳互联信息技术有限公司 Recommendation method, recommendation device, server and storage medium
CN112905879A (en) * 2021-01-29 2021-06-04 北京达佳互联信息技术有限公司 Recommendation method, recommendation device, server and storage medium
CN112882929B (en) * 2021-02-02 2023-08-08 网易（杭州）网络有限公司 Test method, test device, computer equipment and storage medium
CN112882929A (en) * 2021-02-02 2021-06-01 网易（杭州）网络有限公司 Test method, test device, computer equipment and storage medium
CN113515696A (en) * 2021-05-17 2021-10-19 上海众源网络有限公司 Recommendation method and device, electronic equipment and storage medium
CN113468426A (en) * 2021-07-01 2021-10-01 北京明略软件系统有限公司 Information recommendation method and device, electronic equipment and readable storage medium
CN113468426B (en) * 2021-07-01 2024-01-30 北京明略软件系统有限公司 Information recommendation method and device, electronic equipment and readable storage medium
CN113553509A (en) * 2021-07-29 2021-10-26 北京达佳互联信息技术有限公司 Content recommendation method and device, electronic equipment and storage medium
CN113553509B (en) * 2021-07-29 2024-03-01 北京达佳互联信息技术有限公司 Content recommendation method and device, electronic equipment and storage medium
CN116911957A (en) * 2023-09-12 2023-10-20 深圳须弥云图空间科技有限公司 Heterogeneous content mixed recommendation method, device, system and storage medium
CN116911957B (en) * 2023-09-12 2024-01-05 深圳须弥云图空间科技有限公司 Heterogeneous content mixed recommendation method, device, system and storage medium

Similar Documents

Publication Publication Date Title
CN111782957A (en) Method and device for recommending content, computing equipment and storage medium
CN110321422B (en) Method for training model on line, pushing method, device and equipment
CN110941740B (en) Video recommendation method and computer-readable storage medium
JP4524709B2 (en) Information processing apparatus and method, and program
US10949000B2 (en) Sticker recommendation method and apparatus
CN110704739A (en) Resource recommendation method and device and computer storage medium
WO2014143018A1 (en) Efficient and fault-tolerant distributed algorithm for learning latent factor models through matrix factorization
CN104142999A (en) Search result display method and device
CN110413867B (en) Method and system for content recommendation
CN104462594A (en) Method and device for providing user personalized resource message pushing
US20110213669A1 (en) Allocation of Resources
CN108345419A (en) A kind of generation method and device of information recommendation list
US20140278723A1 (en) Methods and systems for predicting workflow preferences
CN112423134B (en) Video content recommendation method and device, computer equipment and storage medium
CN111931055B (en) Object recommendation method, object recommendation device and electronic equipment
CN111259195A (en) Video recommendation method and device, electronic equipment and readable storage medium
US20100125585A1 (en) Conjoint Analysis with Bilinear Regression Models for Segmented Predictive Content Ranking
CN111949807A (en) Data processing method, data processing device, computer equipment and storage medium
CN112905879A (en) Recommendation method, recommendation device, server and storage medium
CN113010790A (en) Content recommendation method, device, server and storage medium
CN111198961A (en) Commodity searching method and device and server
CN112749296B (en) Video recommendation method, device, server and storage medium
WO2020211840A1 (en) Material recommendation method and system
US11100158B1 (en) Fast feature selection for search models
CN113760550A (en) Resource allocation method and resource allocation device

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
REG Reference to a national code

Ref country code: HK

Ref legal event code: DE

Ref document number: 40030683

Country of ref document: HK

SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination